| { | |
| "best_metric": 2.2532453536987305, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.8820286659316428, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.004410143329658214, | |
| "grad_norm": 93.76297760009766, | |
| "learning_rate": 1e-05, | |
| "loss": 13.981, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.004410143329658214, | |
| "eval_loss": 3.361562728881836, | |
| "eval_runtime": 3.9837, | |
| "eval_samples_per_second": 95.89, | |
| "eval_steps_per_second": 24.098, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.008820286659316428, | |
| "grad_norm": 93.67481994628906, | |
| "learning_rate": 2e-05, | |
| "loss": 14.3409, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.013230429988974642, | |
| "grad_norm": 73.69136810302734, | |
| "learning_rate": 3e-05, | |
| "loss": 13.6807, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.017640573318632856, | |
| "grad_norm": 39.784732818603516, | |
| "learning_rate": 4e-05, | |
| "loss": 13.4712, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.022050716648291068, | |
| "grad_norm": 7.986569404602051, | |
| "learning_rate": 5e-05, | |
| "loss": 12.7339, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.026460859977949284, | |
| "grad_norm": 8.91831111907959, | |
| "learning_rate": 6e-05, | |
| "loss": 13.0562, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.030871003307607496, | |
| "grad_norm": 10.288507461547852, | |
| "learning_rate": 7e-05, | |
| "loss": 12.9373, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.03528114663726571, | |
| "grad_norm": 10.245424270629883, | |
| "learning_rate": 8e-05, | |
| "loss": 13.0012, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03969128996692393, | |
| "grad_norm": 8.181429862976074, | |
| "learning_rate": 9e-05, | |
| "loss": 12.7252, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.044101433296582136, | |
| "grad_norm": 9.538784980773926, | |
| "learning_rate": 0.0001, | |
| "loss": 12.5884, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.04851157662624035, | |
| "grad_norm": 10.758712768554688, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 11.8328, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.05292171995589857, | |
| "grad_norm": 8.99941349029541, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 11.8979, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.05733186328555678, | |
| "grad_norm": 7.923396587371826, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 11.6141, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.06174200661521499, | |
| "grad_norm": 8.583168029785156, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 11.7593, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.06615214994487321, | |
| "grad_norm": 6.530113220214844, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 11.5101, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.07056229327453142, | |
| "grad_norm": 5.689020156860352, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 11.6373, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.07497243660418963, | |
| "grad_norm": 4.761807918548584, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 11.1699, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.07938257993384785, | |
| "grad_norm": 4.804562091827393, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 11.3123, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.08379272326350606, | |
| "grad_norm": 4.137475967407227, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 10.9885, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.08820286659316427, | |
| "grad_norm": 3.7448537349700928, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 10.911, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0926130099228225, | |
| "grad_norm": 4.244678497314453, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 11.0157, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0970231532524807, | |
| "grad_norm": 3.664881944656372, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 11.1407, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.10143329658213891, | |
| "grad_norm": 3.5193731784820557, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 10.5017, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.10584343991179714, | |
| "grad_norm": 3.5935089588165283, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 10.6506, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.11025358324145534, | |
| "grad_norm": 3.6575050354003906, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 10.8303, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.11466372657111357, | |
| "grad_norm": 3.4716010093688965, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 10.8933, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.11907386990077178, | |
| "grad_norm": 3.3464410305023193, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 10.7289, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.12348401323042998, | |
| "grad_norm": 3.87205171585083, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 10.7642, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.1278941565600882, | |
| "grad_norm": 3.0769689083099365, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 10.5172, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.13230429988974643, | |
| "grad_norm": 3.3340649604797363, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 10.5238, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.13671444321940462, | |
| "grad_norm": 3.469038963317871, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 10.4521, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.14112458654906285, | |
| "grad_norm": 3.2644364833831787, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 10.5504, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.14553472987872107, | |
| "grad_norm": 3.4640092849731445, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 10.594, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.14994487320837926, | |
| "grad_norm": 3.749213218688965, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 10.8588, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.1543550165380375, | |
| "grad_norm": 2.9205386638641357, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 10.2248, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.1587651598676957, | |
| "grad_norm": 2.901705265045166, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 10.6851, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.1631753031973539, | |
| "grad_norm": 3.4778096675872803, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 10.0603, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.16758544652701213, | |
| "grad_norm": 3.182819366455078, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 10.0709, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.17199558985667035, | |
| "grad_norm": 3.241534471511841, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 10.4049, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.17640573318632854, | |
| "grad_norm": 3.31471848487854, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 10.6116, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.18081587651598677, | |
| "grad_norm": 3.335679531097412, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 10.7684, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.185226019845645, | |
| "grad_norm": 3.6037471294403076, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 10.6011, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.18963616317530319, | |
| "grad_norm": 4.118447303771973, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 10.7639, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.1940463065049614, | |
| "grad_norm": 3.7127935886383057, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 10.3632, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.19845644983461963, | |
| "grad_norm": 4.114322185516357, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 9.9567, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.20286659316427783, | |
| "grad_norm": 3.9941813945770264, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 9.8874, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.20727673649393605, | |
| "grad_norm": 4.080880641937256, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 9.6216, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.21168687982359427, | |
| "grad_norm": 4.799224376678467, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 9.7251, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.2160970231532525, | |
| "grad_norm": 4.749310493469238, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 9.7271, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.2205071664829107, | |
| "grad_norm": 6.439767360687256, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 9.5252, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.2205071664829107, | |
| "eval_loss": 2.4655730724334717, | |
| "eval_runtime": 3.986, | |
| "eval_samples_per_second": 95.835, | |
| "eval_steps_per_second": 24.084, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.2249173098125689, | |
| "grad_norm": 7.9255781173706055, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 10.5467, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.22932745314222713, | |
| "grad_norm": 7.473238468170166, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 10.2634, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.23373759647188533, | |
| "grad_norm": 5.979231834411621, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 10.4718, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.23814773980154355, | |
| "grad_norm": 5.469010353088379, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 10.4348, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.24255788313120177, | |
| "grad_norm": 4.270541667938232, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 10.1428, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.24696802646085997, | |
| "grad_norm": 4.045993328094482, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 10.0831, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.2513781697905182, | |
| "grad_norm": 3.583444595336914, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 10.324, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.2557883131201764, | |
| "grad_norm": 3.483198404312134, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 10.1282, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.26019845644983464, | |
| "grad_norm": 3.4700069427490234, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 10.1232, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.26460859977949286, | |
| "grad_norm": 3.5300607681274414, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 10.1978, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.269018743109151, | |
| "grad_norm": 3.4217031002044678, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 9.9816, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.27342888643880925, | |
| "grad_norm": 3.2927191257476807, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 9.9012, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.27783902976846747, | |
| "grad_norm": 3.17272686958313, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 9.9606, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.2822491730981257, | |
| "grad_norm": 3.1798388957977295, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 10.2441, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.2866593164277839, | |
| "grad_norm": 3.097712278366089, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 9.5621, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.29106945975744214, | |
| "grad_norm": 3.1934633255004883, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 10.0037, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.2954796030871003, | |
| "grad_norm": 2.925327777862549, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 10.0632, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.29988974641675853, | |
| "grad_norm": 2.76534366607666, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 9.8498, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.30429988974641675, | |
| "grad_norm": 2.9064295291900635, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 9.7779, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.308710033076075, | |
| "grad_norm": 3.097909688949585, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 9.9231, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.3131201764057332, | |
| "grad_norm": 2.8864567279815674, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 9.5838, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.3175303197353914, | |
| "grad_norm": 2.7767820358276367, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 9.9516, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.3219404630650496, | |
| "grad_norm": 2.8595311641693115, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 9.6969, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.3263506063947078, | |
| "grad_norm": 2.894693613052368, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 9.8685, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.33076074972436603, | |
| "grad_norm": 2.824695587158203, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 9.8168, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.33517089305402425, | |
| "grad_norm": 2.9962146282196045, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 9.6154, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.3395810363836825, | |
| "grad_norm": 2.707134962081909, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 9.7863, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.3439911797133407, | |
| "grad_norm": 2.5159194469451904, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 9.5093, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.3484013230429989, | |
| "grad_norm": 2.705833911895752, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 9.7085, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.3528114663726571, | |
| "grad_norm": 2.610684871673584, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 9.7547, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.3572216097023153, | |
| "grad_norm": 2.7021446228027344, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 9.7201, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.36163175303197354, | |
| "grad_norm": 2.925966501235962, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 9.6825, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.36604189636163176, | |
| "grad_norm": 2.917950391769409, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 9.7022, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.37045203969129, | |
| "grad_norm": 2.6790931224823, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 9.586, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.3748621830209482, | |
| "grad_norm": 2.8914520740509033, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 9.708, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.37927232635060637, | |
| "grad_norm": 2.735027313232422, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 9.8346, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.3836824696802646, | |
| "grad_norm": 2.8165762424468994, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 9.9393, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.3880926130099228, | |
| "grad_norm": 3.055398941040039, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 9.9979, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.39250275633958104, | |
| "grad_norm": 3.5361881256103516, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 9.853, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.39691289966923926, | |
| "grad_norm": 3.080867290496826, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 9.9319, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.4013230429988975, | |
| "grad_norm": 3.6121015548706055, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 9.6563, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.40573318632855565, | |
| "grad_norm": 3.659212112426758, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 9.8984, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.4101433296582139, | |
| "grad_norm": 3.804407835006714, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 10.0283, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.4145534729878721, | |
| "grad_norm": 4.0184831619262695, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 9.7105, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.4189636163175303, | |
| "grad_norm": 3.6824862957000732, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 9.7631, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.42337375964718854, | |
| "grad_norm": 4.295379638671875, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 9.4494, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.42778390297684676, | |
| "grad_norm": 4.584733963012695, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 9.5438, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.432194046306505, | |
| "grad_norm": 4.690437316894531, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 9.2678, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.43660418963616315, | |
| "grad_norm": 5.595583438873291, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 9.4706, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.4410143329658214, | |
| "grad_norm": 14.302119255065918, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 9.7156, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.4410143329658214, | |
| "eval_loss": 2.3261044025421143, | |
| "eval_runtime": 3.9912, | |
| "eval_samples_per_second": 95.712, | |
| "eval_steps_per_second": 24.053, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.4454244762954796, | |
| "grad_norm": 4.906518459320068, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 9.7847, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.4498346196251378, | |
| "grad_norm": 5.2877984046936035, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 9.873, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.45424476295479604, | |
| "grad_norm": 4.902725696563721, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 9.5166, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.45865490628445427, | |
| "grad_norm": 4.597932815551758, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 9.745, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.46306504961411243, | |
| "grad_norm": 4.229561805725098, | |
| "learning_rate": 5e-05, | |
| "loss": 9.5934, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.46747519294377066, | |
| "grad_norm": 3.610197067260742, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 9.6926, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.4718853362734289, | |
| "grad_norm": 3.2193634510040283, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 9.6004, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.4762954796030871, | |
| "grad_norm": 3.1794707775115967, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 9.5997, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.4807056229327453, | |
| "grad_norm": 3.0589001178741455, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 9.8258, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.48511576626240355, | |
| "grad_norm": 3.0398447513580322, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 9.4736, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.4895259095920617, | |
| "grad_norm": 3.166065216064453, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 9.6422, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.49393605292171994, | |
| "grad_norm": 2.8575689792633057, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 9.2572, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.49834619625137816, | |
| "grad_norm": 3.2390735149383545, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 9.4472, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.5027563395810364, | |
| "grad_norm": 2.7992186546325684, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 9.6051, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.5071664829106945, | |
| "grad_norm": 2.838766574859619, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 9.6754, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.5115766262403528, | |
| "grad_norm": 2.7878000736236572, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 9.4247, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.515986769570011, | |
| "grad_norm": 2.8886728286743164, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 9.7827, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.5203969128996693, | |
| "grad_norm": 2.7461094856262207, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 9.4248, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.5248070562293274, | |
| "grad_norm": 2.7339069843292236, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 9.68, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.5292171995589857, | |
| "grad_norm": 2.7840816974639893, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 9.6279, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.5336273428886439, | |
| "grad_norm": 2.745861530303955, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 9.4962, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.538037486218302, | |
| "grad_norm": 3.0586671829223633, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 9.4334, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.5424476295479603, | |
| "grad_norm": 2.6563634872436523, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 9.6874, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.5468577728776185, | |
| "grad_norm": 2.6909894943237305, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 9.7535, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.5512679162072768, | |
| "grad_norm": 2.728145122528076, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 9.2734, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.5556780595369349, | |
| "grad_norm": 2.8144240379333496, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 9.8901, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.5600882028665931, | |
| "grad_norm": 2.635753631591797, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 9.5175, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.5644983461962514, | |
| "grad_norm": 2.6185126304626465, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 9.4826, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.5689084895259096, | |
| "grad_norm": 2.7537648677825928, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 9.6131, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.5733186328555678, | |
| "grad_norm": 2.5011887550354004, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 9.4247, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.577728776185226, | |
| "grad_norm": 2.6665756702423096, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 9.3427, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.5821389195148843, | |
| "grad_norm": 2.630866050720215, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 9.6966, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.5865490628445424, | |
| "grad_norm": 2.647047996520996, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 9.5611, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.5909592061742006, | |
| "grad_norm": 2.9013044834136963, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 9.3512, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.5953693495038589, | |
| "grad_norm": 2.664992094039917, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 9.1133, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.5997794928335171, | |
| "grad_norm": 2.613198757171631, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 9.7721, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.6041896361631753, | |
| "grad_norm": 2.8741347789764404, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 9.5101, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.6085997794928335, | |
| "grad_norm": 2.7328243255615234, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 9.2628, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.6130099228224918, | |
| "grad_norm": 2.7596280574798584, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 9.477, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.61742006615215, | |
| "grad_norm": 2.807450294494629, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 9.6781, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.6218302094818081, | |
| "grad_norm": 2.744504928588867, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 9.8151, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.6262403528114664, | |
| "grad_norm": 3.2098045349121094, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 9.8578, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.6306504961411246, | |
| "grad_norm": 3.549811363220215, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 9.8097, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.6350606394707828, | |
| "grad_norm": 4.015124797821045, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 9.5808, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.639470782800441, | |
| "grad_norm": 3.831195116043091, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 9.4252, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.6438809261300992, | |
| "grad_norm": 4.39556884765625, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 9.4444, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.6482910694597575, | |
| "grad_norm": 4.465538501739502, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 9.4112, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.6527012127894156, | |
| "grad_norm": 4.611509323120117, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 9.334, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.6571113561190739, | |
| "grad_norm": 5.176872730255127, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 9.3733, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.6615214994487321, | |
| "grad_norm": 8.15030574798584, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 9.1197, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6615214994487321, | |
| "eval_loss": 2.2692179679870605, | |
| "eval_runtime": 3.9787, | |
| "eval_samples_per_second": 96.011, | |
| "eval_steps_per_second": 24.128, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6659316427783903, | |
| "grad_norm": 3.6759140491485596, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 9.4823, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.6703417861080485, | |
| "grad_norm": 3.4207842350006104, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 9.5271, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.6747519294377067, | |
| "grad_norm": 3.703519582748413, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 9.8302, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.679162072767365, | |
| "grad_norm": 3.2789928913116455, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 9.6278, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.6835722160970231, | |
| "grad_norm": 2.9552910327911377, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 9.5895, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.6879823594266814, | |
| "grad_norm": 2.9568581581115723, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 9.5088, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.6923925027563396, | |
| "grad_norm": 2.9085640907287598, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 9.5415, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.6968026460859978, | |
| "grad_norm": 2.9179465770721436, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 9.4631, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.701212789415656, | |
| "grad_norm": 2.654332160949707, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 9.02, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.7056229327453142, | |
| "grad_norm": 2.8916449546813965, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 9.6744, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.7100330760749725, | |
| "grad_norm": 2.5509328842163086, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 9.5784, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.7144432194046306, | |
| "grad_norm": 2.656136989593506, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 9.4054, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.7188533627342889, | |
| "grad_norm": 2.6057004928588867, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 9.543, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.7232635060639471, | |
| "grad_norm": 2.974360942840576, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 9.578, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.7276736493936052, | |
| "grad_norm": 2.5843842029571533, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 9.5427, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.7320837927232635, | |
| "grad_norm": 2.5783538818359375, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 9.4709, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.7364939360529217, | |
| "grad_norm": 2.44590425491333, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 9.4573, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.74090407938258, | |
| "grad_norm": 2.5670814514160156, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 9.5009, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.7453142227122381, | |
| "grad_norm": 2.6164445877075195, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 9.7775, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.7497243660418964, | |
| "grad_norm": 2.614750623703003, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 9.4587, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.7541345093715546, | |
| "grad_norm": 2.6944541931152344, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 9.5405, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.7585446527012127, | |
| "grad_norm": 2.5363221168518066, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 9.4638, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.762954796030871, | |
| "grad_norm": 2.5635743141174316, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 9.5558, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.7673649393605292, | |
| "grad_norm": 2.5865283012390137, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 9.2366, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.7717750826901875, | |
| "grad_norm": 2.468916177749634, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 9.3768, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.7761852260198456, | |
| "grad_norm": 2.4861855506896973, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 9.4156, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.7805953693495039, | |
| "grad_norm": 2.5258901119232178, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 9.2959, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.7850055126791621, | |
| "grad_norm": 2.4535820484161377, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 9.5271, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.7894156560088202, | |
| "grad_norm": 2.4277803897857666, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 9.3665, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.7938257993384785, | |
| "grad_norm": 2.5256242752075195, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 9.341, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.7982359426681367, | |
| "grad_norm": 2.6025824546813965, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 9.8123, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.802646085997795, | |
| "grad_norm": 2.4343557357788086, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 9.4829, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.8070562293274531, | |
| "grad_norm": 2.5988001823425293, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 9.7153, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.8114663726571113, | |
| "grad_norm": 2.562573194503784, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 9.485, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.8158765159867696, | |
| "grad_norm": 2.454115629196167, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 9.3411, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.8202866593164277, | |
| "grad_norm": 2.6099281311035156, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 9.4701, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.824696802646086, | |
| "grad_norm": 2.6781437397003174, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 9.8366, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.8291069459757442, | |
| "grad_norm": 2.4233927726745605, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 9.5856, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.8335170893054025, | |
| "grad_norm": 2.722417116165161, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 9.6465, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.8379272326350606, | |
| "grad_norm": 2.6799838542938232, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 9.821, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.8423373759647188, | |
| "grad_norm": 2.830326795578003, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 9.6341, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.8467475192943771, | |
| "grad_norm": 2.969053030014038, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 9.7395, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.8511576626240352, | |
| "grad_norm": 3.2237770557403564, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 9.9695, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.8555678059536935, | |
| "grad_norm": 3.5367090702056885, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 9.6035, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.8599779492833517, | |
| "grad_norm": 3.7807130813598633, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 9.526, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.86438809261301, | |
| "grad_norm": 3.813859224319458, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 9.2044, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.8687982359426681, | |
| "grad_norm": 4.019887447357178, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 9.0055, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.8732083792723263, | |
| "grad_norm": 4.306025981903076, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 9.0699, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.8776185226019846, | |
| "grad_norm": 5.382304668426514, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 8.8738, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.8820286659316428, | |
| "grad_norm": 6.224758148193359, | |
| "learning_rate": 0.0, | |
| "loss": 8.9482, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.8820286659316428, | |
| "eval_loss": 2.2532453536987305, | |
| "eval_runtime": 3.9946, | |
| "eval_samples_per_second": 95.63, | |
| "eval_steps_per_second": 24.033, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 7284189598580736.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |